As part of an initial survey of our project steering group, we asked for comments on the scoping questions proposed by the ESRC for the seven domains identified by the review. For the data and representation domain, the question was, “How do we live with and trust the algorithms and data analysis used to shape key features of our lives?”
Recognising the vast literature that exists around this subject area, we concentrate, in this blog, only on some of the headline points raised by the experts, who come from seventeen institutions across the world, representing social science, arts, engineering and science disciplines.
Comments from the initial survey did not question the potential for massive advances in computing power and computing skills to enable powerful insights and allow for better decision-making; however, they did reflect a consensus that some areas of this domain are under-researched and not well understood.
They noted the current focus on big data and the commercial possibilities and policy drivers that are informing current discourses about data. This is reflected in the significant amount of funding available to undertake research into big data within academia, within the policy field, and within the commercial sector. There is therefore already much research in place that is exploring big data and its applications. Key findings already suggest that the use of big data is uncertain outside of commercial marketing processes, although the idea of accessing live data in real time has huge potential because of its potential to facilitate almost real-time decision-making. However, an overarching point coming out of the initial survey was the critical role that confidence in these processes plays (i.e. public engagement/involvement with these processes has to be both reflective and safe).
Key areas that the respondents felt required attention in relation to this issue are: the timeliness, representativeness and reliability of data; the potential for bias; and, as things stand, the fact that there is little evidence on exactly how the data might inform policy decisions. Despite the current very positive rhetoric regarding the value of data, experts warned there is an important role to play in urging a more profound examination of the implications – acknowledging the complexity rather than advocating simple provision and acceptance of solutions.
The responses from experts to questions in this domain overlapped significantly with the responses received in relation to the governance and security domain, because much of the promise associated with different forms of data and its analysis are still at an early stage. Some of the more pressing questions demanding attention were stated to be:
- Who owns the data (e.g. governments, technology companies or multinational corporations)?
- Who has control of the data produced?
- Does ownership change when new data sets are formed through the merging of data sets?
- Do the individuals who provided the data have any control or do all internet users end up (unwittingly) becoming the unpaid labour that produces free data for public or private companies?
As a priority, thought must be given to the extent that individuals either have an active role in deciding what to trust or if it is safe to leave this to a third party who acts as a guide or trusted partner in the process. Digital literacy and awareness on how criteria are decided in this form of decision-making were thought to be in short supply, necessitating rigorous quantitative and qualitative investigation to work through all the possible implications.
There was little comment about data representation. This is surprising given the importance of data representation. This of course is not a new issue and attention has already been paid to the value that data analysis and visualisation brings and whether research methodologies change or adapt when studying web based data.
In line with the other domains that form the basis of this study, experts have raised the issue of overlap with other domains; but the emphasis that has been given to the security, privacy, gate-keeping, curation, etc. of data has led some to go further and suggest that the match and overlap with governance and security is so pervasive that the two domains might be usefully merged. We would be interested in your views and any research you are involved in that illustrates commonality or otherwise between data and governance. The Ways of Being Digital website is one of many ways of keeping up to date with progress and letting us know of any relevant research projects that help to answer the ESRC scoping questions.